Strange issue involving XOR with chars

Accelerated Computing CUDA CUDA Programming and Performance
1

I’m attempting to implement RC4 encryption and have run into a bit of an issue. For some reason, CUDA doesn’t seem to be executing the following line:

a[idx*256+y]=a[idx*256+y] ^ S[(S[i] + S[j]) % 255];

Where a and S are both char matrices. I can access a and change all chars, to N for example (a[idx*256+y]=‘N’ ), and it works. It also works fine in emulation mode (I know, I know emulation mode isn’t worth anything).

I’ve been looking for a solution all weekend to no avail. BTW I’m running the latest dev drivers.

Any help would be greatly appreciated

Full Code

__global__ void rc4(char a[N_DATASIZE*N_OBJECTS], string b[N_OBJECTS], int N, size_t *string_size)

{

	int idx = blockIdx.x * blockDim.x + threadIdx.x;

	

	if (idx<N)

	{

		int keysize = string_size[idx];

		

		unsigned char S[256];

		int i, j;

		/* KSA */

		for (i = 0; i < 256; i++)

			S[i] = i;

		unsigned char temp;

		for (i = j = 0; i < 256; i++) {

			j = (j + b[idx][i % keysize] + S[i]) % 255;

			/*SWAP*/

			temp = S[i];

			S[i] = S[j];

			S[j] = temp;

		}

		i = j = 0;

		

		/* PRGA */

		for (int y=0; y<N_DATASIZE; y++)

		{

			i = (i + 1) % 255;

			j = (j + S[i]) % 255;

			/*SWAP*/

			temp = S[i];

			S[i] = S[j];

			S[j] = temp;

			

			a[idx*256+y]=a[idx*256+y] ^ S[(S[i] + S[j]) % 255];

		}

		

	}	

}